Ndless 1.7 for TI-Nspire

[critor]

Just because on TI-BANK and United-TI you can put down (en français vous dîtes "enfoncer" je crois?) members who ask simple questions and miss the answer buried among the previous three hundred messages without worrying if they will feel welcome anymore or not, it doesn't mean you are allowed to do so on Omnimaga.

Please DJ, don't get mad at me, but I'd like to say something.

Not something about your decision. You're right.
It's true that Levak's post seems agressive to me.
Was it meant? I don't know... Just remember that Levak doesn't speak english natively.
He is taking the time to share his knowledge in english, and he may use the right technical terms, but not the right non-technical terms... He might not have wanted to sound so agressive...
Of course, like you I hope there won't be a next time but if there is, I beg you, please take that into consideration before taking a big decision.

No, it's something about your comment.
I don't know about UTI, but on TI-Bank neither Levak nor the TI-Bank team does put down new members... We are taking the time to post comprehensive answers, and repeat them in other topics if necessary.
Thank you for reading.


Now, back on topic.

Thank you all for the Ndless tests and those great news! ^^

[DJ Omnimaga]

Sorry if it sounded like I was saying that you and Levak were putting new TI-BANK members down, it was an example of if you ever decided to do it you may not have any problems, not necessarly saying you were actually doing it. I should probably have said that differently. I'll edit that part of my post to make sure it won't be taken that way anymore. I was saying TI-BANK and United-TI because I do remember people there doing that. I even intervened in one topic to explain that everyone has to start somewhere. Sorry about that.

[critor]

No problem

[calcforth]

I don't think it has anything to do with USB versions.

Actually it kida does. USB 1.x, 2.0 and 3.0 are joined but separate (like oil and water): they have different protocols (if you'll see topology in your device tree list you'll see that USB 1.x devices are connected to UHCI or OHCI hub, USB 2.0 devices to EHCI hub and USB 3.0 devices to xHCI hub), they use different resources, etc. In case of USB 3.0 they even use different wires!

Well, it's nice information to have, but how it's related to our Ndless woes? Well, the answer is simple: since USB 3.0 connector combines three interfaces in one connector it actually takes longer for it to detect the version of USB device is using! And that means that that Ndless have more time to do it's magic before USB hub will start hammering OS with interrrupts (USB 1.x and 2.0 are fully passive - they only generate activity on slave device when hub asks them to do something; USB 3.0 is different but here it's not very relevant because TI-Nspire does not support USB 3.0).

It means that while USB 3.0 per se should not affect Ndless in any way using USB 3.0 connector may help to make the procedure more robust. That's all. Now the only question is: just what is the troubling part of Ndless which makes it impossible to do any work in the presense of the interrupts and can we reduce this "window of vulnerability"? If we can do it then USB 2.0 or may be 1.x connectors should work too.

[ExtendeD]

just what is the troubling part of Ndless which makes it impossible to do any work in the presense of the interrupts and can we reduce this "window of vulnerability"?

I suppose the freeze happens before the code of Ndless is run, we probably won't be able to do much about it.
Since the issue seems to be specific to Levak's configuration, and since the workaround is as simple as unplugging the USB cable before the installation, I won't invest more effort than adding a warning to the ReadMe file.

[Munchor]

I have a doubt with ndless. When you install it, what exactly happens? Like, I can't find a video. But can you calculate, use it as normal.

And when you install, when you turn off anything happens?

Please explain me because I'm quite curious

[bsl]

Got Ndless CAS1.7 running on the emulator
put in syscalls-light_cas-1.7.c:

Code: Select All | Copy To Clipboard

1 2 3 4 5 6 7 8 9

0X102A6CCC         , 0X102A6EC0         , 0X102a6ffc         , 0X102A66C4         , 0X102A481C         , 0X102AC2DC         , 0X102A74D4         , 0X102A31D8

and append to syscalls_cas-1.7.c:

Code: Select All | Copy To Clipboard

1 2 3

, 0X102A49E8         , 0x102a6ffc

You can use Levak's  values given on this thread.
Also, to get it to compile, the Makefile under libndls/
I changed the line:

Code: Select All | Copy To Clipboard

1 2	from: AR = arm-none-eabi-ar to:   AR = arm-elf-ar

[Levak]

I have a doubt with ndless. When you install it, what exactly happens? Like, I can't find a video. But can you calculate, use it as normal.

And when you install, when you turn off anything happens?

Please explain me because I'm quite curious

You can do what ever you want, either work like before, or run asm progs.
Ndless will "just" by-pass the verification of a valid document and "add" the possibilitie of documents that begin with "PRG"
ExtendeD will surely explain better than me.

[Munchor]

You can do what ever you want, either work like before, or run asm progs.
Ndless will "just" by-pass the verification of a valid document and "add" the possibilitie of documents that begin with "PRG"
ExtendeD will surely explain better than me.

Hum. Without ndless I can run Asm progs?

[JosJuice]

You can do what ever you want, either work like before, or run asm progs.
Ndless will "just" by-pass the verification of a valid document and "add" the possibilitie of documents that begin with "PRG"
ExtendeD will surely explain better than me.

Hum. Without ndless I can run Asm progs?

No.

[ExtendeD]

Most fixes suggested have been pushed to the SVN repository.

put in syscalls-light_cas-1.7.c:

The syscalls-*.c files are actually automatically generated by tools/MakeSyscalls from the idc files, I forgot to rebuild them.

Also, to get it to compile, the Makefile under libndls/ I changed the line:

Fixed, thanks.

sleep() isn't setting the timer counter reset value back properly (reading 900D0000 gets its current value, not its reset value)

Do I need to hard-code the OS value? I can't see any port to read the reset value.

[calc84maniac]

sleep() isn't setting the timer counter reset value back properly (reading 900D0000 gets its current value, not its reset value)

Do I need to hard-code the OS value? I can't see any port to read the reset value.

Yeah, there seems to be no way to retrieve the reset value from the hardware (aside from looping while polling the port to get which value comes after 0). Might be best to hard-code the OS value.

[ExtendeD]

Well, ok

I am now releasing binary versions of Ndless 1.7 as a beta release, to get quick feedback before an stable release
-> http://bytecode.fr/miscfiles/ti/nspire/ndless-1.7-beta

Please share the link but avoid distributing copies of the file, since it will be kept up-to-date with the HEAD of the SVN repository.

[jbaem]

wow, finally. thanks man!!!

btw: do u already know how long it's gonna take until the stable release??

[ExtendeD]

I could promote this version to a stable release within a week, if I get enough positive comments on it.